CLINICAL ARTICLE - BRAIN INJURY Comparison of predictability of Marshall and Rotterdam CT scan scoring system in determining early mortality after traumatic brain injury Akhil Deepika 1 & A. R. Prabhuraj 1 & Amrit Saikia 1 & Dhaval Shukla 1 Received: 29 June 2015 /Accepted: 28 August 2015 /Published online: 15 September 2015 # Springer-Verlag Wien 2015 Abstract Background Marshall computed tomographic (CT) classifica- tion is widely used as a predictor of outcome. However, this grading system lacks the following variables, which are found to be useful predictors: subarachnoid/intraventricular hemor- rhage, extradural hematoma, and extent of basal cistern com- pression. A new classification called the Rotterdam grading system, incorporating the above variables, was proposed later. In the original paper, this system was found to have superior discrimination as compared to Marshall grading, however, Rotterdam grading has not been validated widely. We aimed to compare the discriminatory power of both grading systems. Methods This is a prospective study of patients with moderate and severe TBI (Glasgow coma scale (GCS) 3–12) who pre- sented to our casualty. All the patients were followed up for 2 weeks to determine early mortality. The discriminatory pow- er of each grading system was determined using area under the receiver operating characteristic curve (AUC). Results A total of 134 patients, mean age 38.3 (±15.7) years, were recruited for study. The overall mortality was 11.2 %. The mean GCS of these patients was 9.6 (±2.3). There was good correlation between Marshall and Rotterdam grading, r =0.68 (significant at 0.01 level). The Marshall CT classifica- tion had reasonable discrimination (AUC - 0.707), and Rot- terdam grading had good discrimination (AUC - 0.681). Conclusions Both Marshal and Rotterdam grading systems are good in predicting early mortality after moderate and se- vere TBI. As the Rotterdam system also includes additional variables like subarachnoid hemorrhage, it may be preferable, particularly in patients with diffuse injury. Keywords Marshall CT scan classification . Rotterdam CT scan score . Traumatic brain injury Introduction There are several predictors of outcome in patients with trau- matic brain injury (TBI). Among them, computerized tomog- raphy (CT) scan is found very useful for prognosis. The indi- vidual imaging features like basal cistern compression, trau- matic subarachnoid hemorrhage, midline shift, or presence of any abnormality have been found to have class I and II evi- dence for >70 % positive predictability in TBI [2]. Besides individual features, the Marshall CT scan classification system is also found to have good predictability [1, 6]. The variables incorporated in Marshall CT scan classification are given in Table 1. Though the Marshall system is very popular, it has the following limitations. First, the classification of traumatic in- tracranial mass lesion as “evacuated” or “non-evacuated” de- pends on knowledge of what subsequently actually happened to the patient; hence, it can be applied only retrospectively. As patient management could vary between individual neurosur- geons, the hematoma categorization might be difficult to ap- ply prospectively to guide management. Second, the basis of cut-off of 25 cc as volume of mass lesion is not clear. The guidelines for the surgical management of traumatic brain injury mention different cut-offs for specific traumatic lesions [3]. An extradural hematoma (EDH) should be evacuated re- gardless of the Glasgow coma scale (GCS), if it is more than 30 cc. A subdural hematoma (SDH) should be evacuated re- gardless of the GCS if it is more than 10 mm thick or with a midline shift of more than 5 mm. Contusions/parenchymal * Dhaval Shukla neurodhaval@rediffmail.com 1 Department of Neurosurgery, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore 560029, India Acta Neurochir (2015) 157:2033–2038 DOI 10.1007/s00701-015-2575-5 Content courtesy of Springer Nature, terms of use apply. Rights reserved.